Let U⊂C, f:U→C, z∈U. If there exists a C-linear function df∣z∈HomC(C,C) such that for any h∈C, as ∣h∣→0, ∣f(z+h)−f(z)−df∣z(h)∣=o(∣h∣), then f is said to be complex differentiable at point z, and df∣z(1) is called its complex derivative, denoted by f′(z). Note that the complex derivative exists if and only if there exists df∣z∈HomC(C,C) such that
∣h∣→0limhf(z+h)−f(z)−df∣z(h)=0,
i.e.,
h→0limhf(z+h)−f(z)−df∣z(h)=0,
or
h→0limhf(z+h)−f(z)=h→0limhdf∣z(h).
Since dimCC=1, df∣z(h)=f′(z)h, hence
f′(z)=h→0limhf(z+h)−f(z).
Let f(z)=u(Rez,Imz)+iv(Rez,Imz). Set z=x+iy, denote f(x,y):=f(x+iy). Below we will not distinguish f and f. If h approaches z from the real direction, then
f′(z)=∂x∂f=∂x∂u+i∂x∂v,
if h approaches z from the purely imaginary direction, then
Theorem 1.1 Let U⊂C be an open set, z∈U, then f:U→C is differentiable at z if and only if f is differentiable and
∂z∂f=0
i.e.,
∂x∂f=−i∂y∂f
i.e.,
∂x∂u=∂y∂v,∂y∂u=−∂x∂v.
Proof Obviously if f is complex differentiable at z then f is real differentiable at that point, hence the necessity has been shown above.
Conversely, assume f is real differentiable and satisfies the Cauchy-Riemann equations. Then for h∈R2, if ∣h∣→0, then f(z+h)−f=A(h)+o(∣h∣), where
A=(∂x∂uz∂x∂vz∂y∂uz∂y∂vz).
Now we require A(⋅) to be C-linear. Note that the Cauchy-Riemann equations guarantee that A is of the form (ab−ba), so A(⋅) is exactly (∂x∂u+i∂x∂v)(⋅), thus f is complex differentiable at z. □
A complex function is holomorphic if it is complex differentiable at every point of the entire U.
Theorem 1.2 For every power series ∑n=0∞anzn, there exists R∈[0,∞], called the radius of convergence, such that
The series converges absolutely on B(0,R);
The series converges uniformly on compact subsets of B(0,R);
The series diverges on C∖B(0,R), and its terms are unbounded;
d. On B(0,R), the sum of the series is analytic, its derivative is ∑n=1∞nanzn−1, and they have the same radius of convergence;
1/R=n→∞limsupn∣an∣.
Proof If ∣z∣<R, then there exists ρ∈(∣z∣,R), so ρ−1>R−1, hence ∃N∈N such that ∀n≥N, ∣an∣1/n<ρ−1, so ∣an∣<ρ−n, hence ∣anzn∣<(∣z∣/ρ)n. Therefore
n=N∑∞∣an∣∣zn∣<n=N∑∞(ρ∣z∣)n<∞,
so the power series converges absolutely at z. For a given ρ<R, choose ρ0∈(ρ,R), there exists N∈N such that ∀n≥N, ∣an∣1/n<ρ0−1, then similarly, for any ∣z∣≤ρ, for any ϵ>0, there exists N′≥N such that ∀n,m≥N′, we have
j=n∑m∣aj∣∣zj∣<j=n∑m(ρ0ρ)j<ϵ,
so the power series converges uniformly on B(0,ρ), hence it converges uniformly on compact subsets of B(0,R). If ∣z∣>R, then we can find ρ∈(R,∣z∣), so ρ−1<R−1, then for any N∈N there exists n>N such that ∣an∣>ρ−n, so for infinitely many n, ∣anzn∣>(∣z∣/ρ)n, thus the terms of the series are unbounded, and the series diverges.
Since nn→1, ∑n=1∞nanzn−1 has the same radius of convergence. Finally, for ∣z∣<R, set
On the other hand, ∣sn′(z0)−f1(z0)∣→0, so there exists N∈N such that for n≥N, each of the above two terms is less than ϵ/3. For any such n, by the definition of derivative, there exists δ>0 such that ∀z∈B(z0,δ) we have
Theorem 2.3 An arc γ:z=z(t) is rectifiable if and only if its real and imaginary parts are of bounded variation.
Proof Note that ∣x(tj)−x(tj−1)∣≤∣z(tj)−z(tj−1)∣ and ∣y(tj)−y(tj−1)∣≤∣z(tj)−z(tj−1)∣, while ∣z(tj)−z(tj−1)∣≤∣x(tj)−x(tj−1)∣+∣y(tj)−y(tj−1)∣,
thus if x(t),y(t) are of bounded variation then l(γ)<∞, conversely, if l(γ)<∞ then x(t),y(t) are both of bounded variation. □
Theorem 2.4 If γ:[a,b]→C∼R2 is C1, then γ is rectifiable, and l(γ)=∫γds=∫ab∣z′(t)∣dt.
Proof Let P be a partition of [a,b], then ∣z(tj+1)−z(tj)∣=∫tjtj+1z′(t)dt≤∫tjtj+1∣z′(t)∣dt, so
hence l(γ)<∞, γ is rectifiable. Since z′(t) is continuous on [a,b], it is uniformly continuous, so for any ϵ>0, there exists δ>0 such that ∀s,t∈[a,b],∣t−s∣<δ, we have ∣z′(t)−z′(s)∣<ϵ. Take a partition P such that ∀j,∣tj−tj−1∣<δ. Then for any t∈[tj,tj+1], ∣z′(t)−z′(tj)∣<ϵ. Thus
After decomposing mappings into real and imaginary parts, theorems concerning exact forms, closed forms, and curve integrals can be directly applied. If p,q are respectively the partial derivatives of some complex function U with respect to x and y, then the integral ∫γ(pdx+qdy) depends only on the endpoints. Indeed, let p=a+ib,q=c+id, then
pdx+qdy=(a+ib)dx+(c+id)dy=(adx+cdy)+i(bdx+ddy),
if we can find V,W such that dV=adx+cdy,dW=bdx+ddy, then U:=V+iW satisfies dU=pdx+qdy. Conversely, if there exists a complex-valued function U such that dU=pdx+qdy, then set U=V+iW, we have
∂x∂U=adx+ibdx=∂x∂V+i∂x∂W,
giving ∂V/∂x=a,∂W/∂x=b, similarly, ∂V/∂y=c,∂W/∂y=c, so adx+cdy,bdx+ddy are each exact, hence the integral ∫γ(pdx+qdy) is independent of the path.
If the integral ∫γfdz is independent of the specific path, i.e., fdz is exact, then there exists F(z) such that dF=fdx+ifdy, so
∂x∂F=f,∂y∂F=if,
thus F(z) satisfies the Cauchy-Riemann equations.
From the above discussions, it can already be concluded that since dz, zdz are exact, or more generally, zndz(n≥0) are all exact, as long as the integral is closed, we have
∮dz=∮zdz=∮zndz=0.
Secondly, if the region is simply connected, f is holomorphic and its partial derivatives are continuous, then ∮fdz=0. This is because by the Cauchy-Riemann equations, we have respectively
This is the proof initially given by Cauchy. In the corollary of Poincaré’s lemma, if U is simply connected, then closed is equivalent to exact on U, but the proof of Poincaré’s lemma requires the partial derivatives of f to be continuous. Later, Goursat found that the condition of continuous partial derivatives is unnecessary, thus giving the following first version of the Cauchy integral theorem.
Theorem 2.5 (Goursat) Let R be a rectangle, f holomorphic on R, then
∫∂Rfdz=0.
Proof For a region H, denote η(H):=∫∂Hfdz. For R, divide it into four congruent rectangles R11,R21,R31,R41, then
η(R)=η(R1)+η(R2)+η(R3)+η(R4).
Thus at least one Rj11 satisfies ∣η(Rj1)∣≥4−1∣η(R)∣. Continue dividing this Rj11, there exists Rj22 satisfying ∣η(Rj22)∣≥4−1∣η(Rj22)∣. Continuing this way, we obtain a descending sequence of rectangles {Rjkk}k=1∞, where ∣η(Rjkk)∣≥4−k∣η(R)∣. By the Cantor intersection theorem, ⋂k=1∞Rjkk=z0∈R. Since R is complex differentiable at z0, for any ϵ>0, there exists δ>0 such that
Let the diagonal of the original rectangle be d, perimeter be L, then the diagonal of Rn is dn=2−nd, perimeter Ln=2−nL, so
4−n∣η(R)∣≤∣η(Rn)∣<ϵ∫∂Rndnds=ϵdnLn=4−nϵdL,
i.e., ∣η(R)∣<ϵdL. Since ϵ is arbitrary, we get ∫∂Rfdz=η(R)=0. □
Theorem 2.6 Let R be a rectangle, with finitely many points {ζj}1n in its interior, f holomorphic on R∖{ζj}. If
z→ζjlim(z−ζj)f(z)=0,∀j=1,2,⋯,n,
then
∫∂Rfdz=0.
Proof Decompose R into small rectangles so that each contains at most one ζj. Thus the problem reduces to the case of a single point set {ζj}={ζ}. In this case, divide R into nine small rectangles, with only the central one R0 containing ζ. Apply the first version of Cauchy’s integral theorem to the other eight rectangles, then sum all integrals to get
∫∂Rfdz=∫∂R0fdz.
For any ϵ, by the theorem condition, we can choose R0 sufficiently small such that on ∂R, ∣f(z)∣<ϵ/∣z−ζ∣. Since such R0 can be chosen arbitrarily, assume R0 is a square with ζ at its center. Then
∫∂R0fdz≤ϵ∫∣z−ζ∣∣dz∣.
Now estimate the integral on the right. The integral is symmetric, so we only need to compute one side. Suppose R0 has side length 2s, after translation ζ=0, consider the top side of the rectangle, then ∣dz∣=dx, ∣z∣≥s. So the integral on the top side is less than ∫−sssdx=2, overall
∫∂R0∣z−ζ∣∣dz∣<8,∫∂R0fdz<8ϵ.
Since ϵ is arbitrary, the proposition follows. □
With the Cauchy integral theorem for rectangles, we can prove the Cauchy integral theorem for disks without requiring the partial derivatives of f to be continuous.
Theorem 2.7 Let f be holomorphic on an open disk D, then for every closed curve γ in D, we have
∫γfdz=0.
Proof Let the center be x0+iy0. For x+iy∈D, define σ as the path from x0+iy0 horizontally to x+iy0, then vertically to x+iy. By Goursat’s theorem, if τ is the path first vertically then horizontally, then the integrals of f along these two paths are equal. Set
F(z)=∫σfdz=∫τfdz,
considering the integrals along τ and σ with respect to partial x and partial y respectively, we find
∂x∂Fz=f(z),∂y∂Fz=if(z),
so fdz is exact, hence the integral of f over any closed curve is zero. □
Corollary 2.8 Let D be an open disk with finitely many points {ζj}1n in its interior. Let f be holomorphic on D, and
z→ζjlim(z−ζj)f(z)=0,∀j=1,2,⋯,n,
then for every closed curve γ in D, we have
∫γfdz=0.
Proof Replace the Goursat theorem used in the proof of the previous theorem with its version with punctured points. □
Since x(t)2+y(t)2=1, we have 2x(t)x′(t)+y(t)y′(t)=0, i.e., xx′+yy′=0. x,y or x′,y′ obviously cannot be identically zero, the only possibility is that there exists k(t) such that x′=−ky,y′=kx. Thus
φ′(t)=xy′−yx′=k(x2+y2)=k(t),
so x′=−φ′y,y′=φ′x. Substituting yields
F′≡0.
This shows F is constant. Note that at t=0
F=F(0)=(x(0)−cos(φ0))2+(y(0)−sin(φ0))2=0,
so F is identically zero on [0,b]. This proves cos(φ(t))=x(t),sin(φ(t))=y(t). □
Definition 3.3 Let c:[0,b]→S1 be a closed curve, φ its angle function. Define
n(c):=2π1(φ(b)−φ(0)),
called the winding number of c. Since c is closed, by Lemma 3.2, n(c)∈Z.
Now we extend the concept of winding number to general closed curves.
Definition 3.4 Let p∈R2, γ:[0,b]→R2∖{p}, set d(t):=γ(t)−p,ρ(t):=∣d(t)∣,c(t):=d(t)/ρ(t), then d,ρ,c∈C1, and γ(t)=p+ρ(t)c(t). Define
n(γ,p):=n(c)
as the winding number of γ about p.
Proposition 3.5 Let γ(t):=p+ρ(t)c(t) be a closed curve [0,b]→R2∖{p}. Set (x(t),y(t)):=γ(t)−p, then
so ω0 is closed. Integrals of closed forms along freely homotopic paths are always equal, so by Proposition 3.5, necessity is proven.
Sufficiency: Let γj=p+ρjcj, let φj be the angle function of cj, set φ(t,τ):=(1−τ)φ0(t)+τφ1(t), H(t,τ):=p+(cos(φ(t,τ)),sin(φ(t,τ))), and set γj:=p+cj and
Hj(t,τ):=p+(1−τ)+τρj(t)ρj(t)cj(t).
Obviously Hj is a free homotopy from γj to γj. If n(γ0,p)=n(γ1,p), then ∀τ∈[0,1] we have
so H is a free homotopy from γ0 to γ1. In summary
γ0≃γ0≃γ1≃γ1,
proving sufficiency. □
Generally, let γ:[a,b]→C be a closed rectifiable curve, w∈/im(γ), define
u:[a,b]t⟶S1=∂D⊂C⟼∣γ(t)−w∣γ(t),
then by Theorem 3.6, if γ is once differentiable then n(γ,w)=n(u). Now we extend the winding number to rectifiable curves and express it using complex integrals.
Lemma 3.7 (Special case of Lifting Lemma) Let π:R→S1,s↦ei2πs be the universal cover. Then there exists a lift u~:[a,b]→R of u such that π∘u~=u.
Proof If u is once differentiable then u~ can be taken as φ(t)/2π, which also proves existence for piecewise differentiable curves. Since rectifiable curves can be approximated by piecewise differentiable curves, the lemma follows. □
Definition 3.8 Let γ:[a,b]→C, w∈/im(γ), u defined as above, u~ its lift. Define the winding number of γ about w as n(γ,w)=u~(b)−u~(a), an extension of Definition 3.3.
Theorem 3.9 The winding number varies continuously under free homotopy, i.e., let γj:[a,b]→C, w∈/im(γj) rectifiable, H(t,τ) a free homotopy from γ1 to γ2, then n(τ):=n(H(⋅,τ),w) is continuous. In particular, if γ1,γ2 are freely homotopic, then n(γ1,w)=n(γ2,w).
Proof Apply the lifting lemma to H to get H~, then n(τ)=H~(b,τ)−H~(a,τ). Since H~ is continuous, n(τ) is naturally continuous. Since n(τ) always takes integer values, we have n(γ1,w)=n(0)=n(1)=n(γ2,w). □
Theorem 3.10 Let γ:[a,b]→C be a rectifiable curve, w∈/im(γ). Then
n(γ,w)=2πi1∫γz−w1dz.
Remark Rectifiable curves can be uniformly approximated by piecewise differentiable curves, so it suffices to prove the proposition for the C1 case. Two proofs are given here.
Let f(z) be holomorphic on an open disk, γ a closed curve inside it. For a point a∈Δ outside γ, consider the function
F(z):=z−af(z)−f(a),
this function is holomorphic on D∖{a}, so as z→a, (z−a)F(z)=f(z)−f(a)→0. Then applying the Cauchy integral theorem gives
∫γz−af(z)−f(a)dz=0,
i.e.,
∫γz−af(z)dz=f(a)∫γz−adz.
Note that the integral on the right is precisely 2πi⋅n(γ,a). Generally, we have the following theorem.
Theorem 4.1 Let f be holomorphic on an open disk D, γ a closed curve in D, a∈C∖im(γ), then
n(γ,a)f(a)=2πi1∫γz−af(z)dz.
Proof The case a∈D has been argued above. If a∈/D, then the winding number n(γ,a) and the integral on the right are also zero, so the formula still holds. □
For a fixed closed curve γ, for those points z with winding number n(γ,z)=1, we have
f(z)=2πi1∫γw−zf(w)dw,n(γ,z)=1.
This is the Cauchy integral formula. In particular, for an open disk D⊂C, let f:D→C be holomorphic, then taking a slightly larger open disk on which f is also holomorphic yields
Lemma 4.2 Let ϕ(w) be a continuous function on an arc γ, then the function
Φn(z):=∫γ(w−z)nϕ(w)dw
is analytic on C∖im(γ), and Φn′(z)=nΦn+1(z).
Proof First prove Φ1 is continuous. Let z0 be a point not on γ, take B(z0,δ) not intersecting γ, then for all z∈B(z0,δ/2),w∈γ, we have ∣w−z∣>δ/2. Thus from
Now prove by induction that Φn is continuous. Assume the proposition holds for n−1, then taking ψ:=ϕ(w)/(w−z0) and applying the induction hypothesis shows the function
Definition 5.1 (Chain, Cycle) Let U⊂C be an open set, R the set of all rectifiable curves on U, denote C0(U):=Z⊕U,C1(U):=Z⊕R, abbreviated as C0,C1. A 1-chain is a formal sum γ=∑j=1najγj∈C1. Denote im(γ):=⋃1nim(γj). For any f∈C(im(γ)), define
∫γfdz:=j=1∑naj∫γjfdz.
Let ι:U→C0 be the canonical inclusion map. Define
where γj:[αj,βj]→C. Elements in ker(∂) are called cycles. For a cycle γ and z∈/im(γ), define
n(γ,z):=2πi1∫γw−zdw.
Definition 5.2 Let U⊂C be an open set, γ∈C1(U) a cycle. γ is said to be homologous to 0 in U if ∀z∈C∖U, n(γ,z)=0.
In the following, we may assume the cycle γ is a sum of closed curves.
Lemma 5.3 Let U⊂C be a simply connected open set, then any cycle on U is homologous to 0.
Proof Each γj is homotopic to a point, and homotopy preserves winding number. □
Theorem 5.4 (General Form of Cauchy Integral Theorem) Let U⊂C be an open set, γ a cycle on U homologous to 0. Let f be a holomorphic function on U. Then
∫γfdz=0.
Proof First assume U is bounded. For any δ>0, take n∈Z such that 2−n<δ, consider the family of closed squares Qn of side length 2−n in the complex plane. Let {Qj}j∈J⊂Qn be those closed squares entirely contained in U, since U is bounded, it is indeed a finite set. Choose δ sufficiently small so that {Qj} is nonempty. Consider the cycle
Γn:=j∈J∑∂Qj.
Set
Un:=int(j∈J⋃Qj),Γn′:=∂Un,Vn:=j∈J⋃(intQj).
Let γ be a cycle on U homologous to 0. Choose δ sufficiently small such that im(γ)⊂Un. Let ξ∈U∖Un, then there exists Q∈Q∖{Qj} such that ξ∈Q and Q∈/U. Pick ξ0∈Q∖U. Since γ is homologous to 0, n(γ,ξ0)=0. We can connect ξ and ξ0 by a straight line L⊂Q such that L∩Un=∅, then L⊂C∖im(γ), meaning ξ0,ξ are in the same connected component of C∖im(γ). Since the winding number is continuous in the point, n(γ,ξ)=n(γ,ξ0)=0. Since Γn′⊂U∖Un, for any ξ∈Γn′, n(γ,ξ)=0.
Let z∈Vn, then there exists a unique j0∈J such that z∈Qj0. By the Cauchy integral formula for rectangles, we have
If U is unbounded, take R sufficiently large such that im(γ)⊂D(0,R) and set U′:=U∩D(0,R).
□
Corollary 5.5 Let U⊂C be a simply connected open set, γ a cycle in U, f a holomorphic function on U. Then
∫γfdz=0.
Theorem 5.6 (General Form of Cauchy Integral Formula) Let U⊂C be an open set, γ a cycle on U homologous to 0. Let f be a holomorphic function on U, a∈U∖im(γ). Then
Corollary 6.2 (Taylor Expansion) Let D:=D(z0,r)⊂C be a disk, f:D→C holomorphic, then f equals a power series on D, in particular
f(z)=2πi1n=0∑∞(∫∂D(w−z0)n+1f(w)dw)(z−z0)n.
Comparing the coefficients from Theorem 6.1 and the Cauchy integral formula with the Taylor series also yields the derivative formula
f(n)(z)=2πin!∫∂D(ζ−z)n+1f(ζ)dζ,
and this method is simpler than the one we used in Section 4.
Theorem 6.3 (Morera) Let U⊂C be an open set, f:U→C continuous. If for every rectifiable closed curve γ in U, ∫γfdz=0, then f is holomorphic.
Proof The condition implies fdz is exact. □
Theorem 6.4 Let U⊂C be an open set, {fn} a sequence of holomorphic functions on U, converging uniformly to f on every compact subset of U. Then f is also a holomorphic function on U.
Proof For any D⊂U, for z∈D, fn(z)=2πi1∫∂Dw−zfn(w)dw, letting n→∞ gives f(z)=2πi1∫∂Dw−zf(w)dw. By Theorem 6.1, f is holomorphic on D. Thus f is holomorphic on the entire U. □
Cauchy Estimates, Liouville’s Theorem, Fundamental Theorem of Algebra#
Theorem 6.5 (Cauchy’s estimate) Let D:=B(z0,r)⊂C be a disk, f:D→C holomorphic. Set M:=maxz∈∂D∣f(z)∣. Then
Theorem 6.6 (Liouville) Let f:C→C be a bounded holomorphic function, then f is constant.
Proof Suppose ∣f∣≤M, let z0∈C, for any R∈R apply Cauchy’s estimate on B(z0,R) to get ∣f′(z0)∣≤MR−n, since R is arbitrary, f′(z0)=0, so f′≡0. Hence f is constant. □
Definition 6.7 If a function f is holomorphic on the entire complex plane, it is called an entire function.
Corollary 6.8 If the real or imaginary part of an entire function f is bounded, then f is constant.
Proof Let f=u+iv. Suppose u is bounded. Consider g:=exp(f), then ∣g∣=∣eu∣. Since u is bounded, so is g, thus by Liouville’s theorem g=exp(f) is constant, so f is constant. □
Theorem 6.9 (Fundamental Theorem of Algebra)C is algebraically closed: Let f∈C[X], degf≥1. Then f has at least one root in C.
Proof Assume f has no root in C, then 1/f is holomorphic on all of C. Since deg≥1, as ∣z∣→∞, ∣f∣→∞, so 1/f→0, hence 1/f is bounded. By Liouville’s theorem, 1/f is constant, so f is also constant, contradicting degf≥1. Therefore f has at least one root in C. □
Lemma 6.10 Let U⊂C be a connected open set, f a holomorphic function on U, z0∈U. Suppose for all n≥1, f(n)(z0)=0. Then f≡f(z0) on U.
Proof Set
V:={z∈U:f(z)=f(z0),f(n)(z)=0,∀n≥1}.
Since z0∈V, V is nonempty. Clearly V is closed. Let z∈V, then there exists D(z,ϵ)⊂U such that f is holomorphic on D(z,ϵ), so f can be expanded as a power series in D and all coefficients of nonzero terms are zero, hence f is identically f(z0) on D, so all derivatives of f are zero on D, thus D⊂V. This shows V is open. Since U is connected, V=U. □
Lemma 6.11 Zeros of holomorphic functions are isolated: Let U⊂C be a connected open set, f a non-constant holomorphic function on U, z0∈U, then there exists r>0 such that z0 is the unique zero of f−f(z0) in D(z0,r).
Proof By Lemma 6.10, there exist r0>0 and m∈N≥1 such that on D′:=D(z0,r0)
f(z)=f(z0)+am(z−z0)m+⋯,am=0,
then on D′, f−f(z0)=(z−z0)mg(z), where the holomorphic function g satisfies g(z0)=am≥0. Since g is continuous, there exists 0<r<r0 such that on D:=D(z0,r), g(z)=0. Thus z0 is the unique zero of f−f(z0) on D. □
Corollary 6.12 Let U⊂C be a connected open set, f,g holomorphic functions on U. If there exists a sequence of points {zn} such that zn→z∈U and f(zj)=g(zj),∀j∈N, then f≡g on U.
Proof By continuity f(z0)=g(z0), then apply Lemma 6.11. □
Theorem 6.13 (Logarithm) Let U⊂C be a simply connected open set, f a holomorphic function on U, with f=0 pointwise. Then there exists a holomorphic function g on U such that eg=f.
Prooff′/f is holomorphic on U. For a∈U, choose α∈C such that eα=f(a). For any z∈U, let γ be a rectifiable curve connecting a and z, define
g(z):=α+∫γff′dz,
then g is independent of the choice of γ, so g is holomorphic and g′=f′/f. Set h:=exp(−g)f, then h(a)=1. Further
h′(z)=−e−g(z)g′(z)f(z)+f′(z)e−g(z)=0,∀z∈U.
Thus h≡1. Therefore eg=f on U. □
Theorem 6.14 (Taking Roots) Let U⊂C be a simply connected open set, f a holomorphic function on U, with f=0 pointwise. Then there exists a holomorphic function g on U such that gn=f.
Proof There exists a holomorphic function h such that eh=f. Set g:=eh/n, then gn=f. □
Open Mapping Theorem, Biholomorphic Mappings, Local Normal Form#
Definition 6.15 (Biholomorphic function) Let U,V be two open sets in C. A holomorphic function f:U→V is called biholomorphic if f is bijective and f−1V→U is holomorphic.
By the inverse function theorem, if f:U→C is holomorphic, z0∈U, and f′(z0)=0, then f is locally biholomorphic at z0.
Theorem 6.16 (Local Normal Form of Holomorphic Functions) Let f be holomorphic on an open set U⊂C, z0∈U, f(m)(z0)=0 and ∀j≤m−1,f(j)(z0)=0. Then there exist D:=D(z0,r)⊂U and a biholomorphic function h on D such that on D, f−f(z0)=hm.
Proof Locally on D′:=D′(z0,r) we can write f−f(z0)=(z−z0)mg(z), where g is holomorphic, g(z0)=am=0. Let h~ be a holomorphic function on D′ satisfying h~m=g, h:=(z−z0)h~, then h is holomorphic on D′ and f−f(z0)=hm. Note that h′(z0)=h~(z0)=0, so by the inverse function theorem, there exists D⊂D′ such that h is biholomorphic on D. □
Lemma 6.17gm(z):C→C,z↦zm is a proper and open map; further, if z=0, then gm−1(z) contains exactly m points.
Proof It is easy to see gm(z) is proper. Let z=reiθ, consider the equation wn=z. Let w=ρeiψ, then ρm=r, mψ=θ+2kπ. The real number ρ=r1/m is unique, while ϕ has exactly m solutions modulo 2π. This proves gm−1(z) has m points. □
Theorem 6.18 (Open Mapping Theorem) Let f be a non-constant holomorphic function on a connected open set U⊂C, then f is an open map.
Proof It suffices to show that for any z0∈U, there exists r>0 such that for any r′<r, D′=D(z0,r′), f(D′) is open. Indeed, by Theorem 6.14, there exist m≥1, a biholomorphic function h, and a disk D=:D(z0,r) such that on D, f−f(z0)=hm. By Lemma 6.15, z↦zm is open, so such D is as desired. □
Theorem 6.19 Let f be a holomorphic function on U. If f is injective, then f:U→f(U) is biholomorphic.
Proof It suffices to prove the proposition for each connected component of U. Below assume U is connected. f(U) is open. ∀z0∈U, f′(z0)=0, otherwise by Theorem 6.14, f is at least m(m≥2)-to-one around z0, so by the inverse function theorem, f−1 is locally holomorphic at f(z0). Since z0 is arbitrary, f is biholomorphic. □
Theorem 6.20 (Maximum Modulus Principle) Let U⊂C be a connected open set, f a non-constant holomorphic function on U. Then there does not exist z0∈U such that ∣f(z0)∣=supz∈U∣f(z)∣.
Proof This is because f(U) is open. □
Considering neighborhoods U′⊂U for any z0∈U in the maximum modulus principle, it further follows that f cannot attain a local maximum at any z0∈U. Similarly, we have the dual “minimum modulus principle”.
Proposition 6.21 Let U⊂C be a connected open set, f a non-constant holomorphic function on U, with f=0 pointwise. Then there does not exist z0∈U such that ∣f(z0)∣=infz∈U∣f(z)∣.
Corollary 6.22 Let U⊂C be a bounded connected open set, f a non-constant holomorphic function on U. Then the maximum of ∣f∣ on U is attained on ∂U. If further f is nonvanishing pointwise, then the minimum of ∣f∣ is attained on ∂U.
Let z0∈C. A Laurent series centered at z0 is a series of the form:
n=1∑∞bn(z−z0)−n+n=0∑∞an(z−z0)n,
where an,bn∈C. ∑n=1∞bn(z−z0)−n is called the principal part, ∑n=0∞an(z−z0)n the regular part.
Let f:=∑n=1∞bn(z−z0)−n+∑n=0∞an(z−z0)n, R the radius of convergence of its regular part. For the principal part, we can also define a “radius of divergence” r. Indeed, set S:=∑j=0∞bjzj, consider the radius of convergence R′ of S, then the principal part converges if and only if S((z−z0)−1) converges. Thus if ∣(z−z0)−1∣>R′ i.e., ∣z−z0∣<1/R′, the principal part does not converge, conversely, if ∣z−z0∣>1/R′, the principal part converges. So we can set r=1/R′. Define the open annulus
D(z0,r,R):={z∈C:r<∣z−z0∣<R},
then f converges uniformly on compact subsets of D(z0,r,R). Since each partial sum is holomorphic, f is holomorphic on D(z0,r,R).
Lemma 7.1 Let g be a holomorphic function on D(z0,r,R), r<r1<r2<R. Then
∫∂D(z0,r1)g(z)dz=∫∂D(z0,r2)g(z)dz.
Proof Set γ:=∂D(z0,r1)−∂D(z0,r2). Since ∂D(z0,rj) are homotopic, γ is homologous to 0 in D(z0,r,R), so by the general form of Cauchy’s integral theorem, ∫γgdz=0, proving the lemma. □
Theorem 7.2 Let f=∑n=−∞∞an(z−z0)n be a Laurent series convergent on D(z0,r,R), r<ρ<R. Then
Theorem 7.3 (Laurent Expansion) Let f be a holomorphic function on D(z0,r,R), then f can be represented as a Laurent series centered at z0 on D(z0,r,R).
Proof By Theorem 7.2, the coefficients of the Laurent series, if they exist, are unique. It suffices to show that f can be expanded as a Laurent series on each D(z0,r′,R′),r<r′<R′<R. For each z∈D(z0,r′,R′), n(∂D(z0,R′)−∂D(z0,r′),z)≡1. Then by the general form of Cauchy’s integral theorem,